Structural Racism and Data Center Siting: Environmental Injustice through Artificial Intelligence Expansion

This paper proposes a research study, with a primary guiding question and scaffolding questions within each other. How does the placement and operations of data centers in historically marginalized communities shape environmental conditions, health outcomes, and social equity over time? 

• In what ways do structural racism and digital redlining influence the geographic distribution of data centers, and how do these placement contribute to cumulative disadvantage and racialized health outcomes? 

• How do community perceptions of trust, collective efficacy, and lived experience act as a protective factor between data centers, environmental burdens, and long-term health outcomes. 

The accelerating demand for artificial intelligence and purposeful placement of data centers replicate and deepen historical legacies of structural racism and redlining in the United States. They concentrate pollution, heat, and noise in predominately black or low-income neighborhoods, contributing to cumulative environmental exposures and worsened health outcomes across the life course.

Proposed Data and Methods: Quantitative and Qualitative Backing

This research project will approach the topic with a mixed method, longitudinal approach to link environmental changes from a data center to community health and social outcomes. This will combine quantitative designs with community centered qualitative methods to explore both measured exposure and lived experiences. Some of the most identifying information in the environment to the future health outcomes of its community is the temperature, air quality, level of air pollution, noise pollution, and water quality (Thamman et al., 2025; Woo et al., 2019). These are all factors that change when a data center is built and operated in a new community, or when previous infrastructure is converted into a facility to run artificial intelligence programing out of (Guidi et al., 2024). To map this, the most ideal form of collection would occur before and after the placement of a data center, with a long period of time to truly see the outcomes. 

Qualitative collection of data would help round out and humanize the data. Another phase of this research would include surveying the local community and trying to gauge their self-rated health, feelings of collective efficacy, and general opinions on their local data center. These three things help gauge the relationship between the data center and the people who live there, along with the health outcomes related. This would require going into the community and finding individuals who give their consent to give their answers. Some background and/or demographic information that would be helpful to collect as well would be proximity to data center, Voting registration and racial/ethnic demographics. Following is the connection each factor has to the literature. 

Self-Rated Health

In Appendix B, the potential survey or interview questions, the first two questions (B.1 and B.2) correspond to the individual’s self-rated health. This ties back to the first section of the review of literature, which discusses the connection of cumulative advantage or disadvantage to an individual’s self-rated health in adulthood. There is potential that data centers may have adverse, long-term health effects for those who currently live nearby. The literature shows that there is current health issues arising from data centers (Booker, 2025; Guidi et al., 2024). Not only that, but the environmental consequences of data centers parallel other, more general environmental health issues (Abi Deivanayagam et al., 2023; Alvarez, 2023; Barberian et al., 2022). 

General Opinions on Local Data Center

Questions B.5, B.6, B.7, and B.16 all touch on the location and presence of a data center for the respondents. This question hits on two different things within data centers, if there is one nearby and if it bothers them in any way. While data centers are being found to be detrimental and have adverse health outcomes, one cannot assume that someone also dislikes it being in their community. Effective community action does not come from an outsider coming in and telling them what they assume the issue to be, and collecting the opinions of the data center will help gauge this. This section also serves to collect demographic information for those near a data center and its proximity to their household and community.

Voting Registration

Voting registration will help gauge the amount of civic engagement and participation a community and individual has. While this does focus more on local level engagement, there is a connection between being registered to vote and being involved in local elections or civic issues. On the anterior of this, self-rated health has a connection to voter turnout as well (Mattila et al., 2013). This connects to B.11 and B.15, commenting on local engagement and not only if they are registered to vote, but if their records are up to date in their community.

Racial/Ethnic Demographics

The collection of racial and ethnic demographic data (connected to B.13) is essential to understanding the disproportionate burden of data centers in marginalized communities. This is highlighted in the first part of the review of literature, which highlights that structural racism and historical housing policies have centralized environmental hazards in communities of color (Smith et al., 2022; Booker, 2025). The gathering of this information will directly connect the placement of data centers to racialized health outcomes and cumulative disadvantage across the life course (Kravitz-Wirtz, 2016; Burgard et al., 2021).

This proposal employs a mixed-method design that integrated quantitative environmental data collection with qualitative community-centered data that captures the experience as a whole. The quantitative sources include publicly available datasets on air quality, temperature, water quality, and nosing pollution, with census and quality data on income, education, and health coverage. These can help place the before and after environmental effect of a data center in a community. Qualitative sources complement this by gathering self-rated health, perceptions of environmental change, and collective efficacy through surveys and interviews.

References (for Web Page, full in attached PDF)

Abi Deivanayagam, T., English, S., Hickel, J., Bonifacio, J., Guinto, R. R., Hill, K. X., ... & Devakumar, D. (2023). Envisioning environmental equity: climate change, health, and racial justice. The Lancet, 402(10395), 64-78.

Alvarez, C. H. (2023). Structural racism as an environmental justice issue: a multilevel analysis of the state racism index and environmental health risk from air toxics. Journal of racial and ethnic health disparities, 10(1), 244-258.

Booker, M. D. (2025). Digital redlining: AI infrastructure and environmental racism in Contemporary America. 

Guidi, G., Dominici, F., Gilmour, J., Butler, K., Bell, E., Delaney, S., & Bargagli-Stoffi, F. J. (2024). Environmental burden of United States data centers in the artificial intelligence era. arXiv preprint arXiv:2411.09786. 

Liu, Z., Yang, L., & Wang, X. (2024). Neighbourhood social capital and political participation in neighbourhood governance: The case of Beijing, China. Transactions in Planning and Urban Research, 3(1-2), 82-96. 

Thamman, R., Nasser, S. A., Ferdinand, K. C., Al-Kindi, S., & Brandt, E. (2025). Exposing Inequality: Environmental Injustice and Cardiovascular Health Disparities. JACC: Advances, 4(7), 101875.

Woo, B., Kravitz-Wirtz, N., Sass, V., Crowder, K., Teixeira, S., & Takeuchi, D. T. (2019). Residential segregation and racial/ethnic disparities in ambient air pollution. Race and social problems, 11(1), 60-67.